Antisense transcription signals relative to sense strand expression from Arabidopsis whole genome tiling arrays^a.

Table 2. Footnotes. a: Gene annotation is from TAIR Release 9 (http://www.arabidopsis.org/). Arabidopsis whole genome tiling array data was from previous reports [82], [83]. For each gene, the ratio of sense/antisense exon signal is calculated according to the following formula: ratio = [(sense exon signals/probe numbers)/(antiense exon signals/probe numbers)]/[(sense intron signals/probe numbers)/(antiense intron signals/probe numbers)]. See Supplemental Text File S1 and Datafile S4 for details. b: One-tailed binomial distribution, normal approximation model, except as noted. c: Validated and predicted miRNA targets were extracted from ASRP database for miRNAs 156, 162, 163, 168, 172, 393, 400, 403, 472, 773 and 780 (http://asrp.cgrb.oregonstate.edu). These targets produce significant numbers of antisense siRNAs [10], [12]–[14];. d: Genes reported as “unknown” were collected from the TAIR9 release for Arabidopsis genome (http://www.arabidopsis.org). e: Small open reading frames (sORFs) were from [27]. f: Genes with antisense transcript units were from [58]. g: Genes with antisense transcripts verified by quantitative RT-PCR were from Y. Xiao and C.D. Town, personal communication. h: Unknown genes with different confidence ratings were from TAIR9 (http://www.arabidopsis.org). Zero rating means no expression data. One star rating means there is weak EST data, and/or another type of low quality functional evidence. Higher (2–5 star) rankings derive from qualitative meta-analysis of full-length cDNAs, proteomics, moncot and dicot cross-species sequence alignments, and genomic conservation. i: Protein-coding genes with antisense smRNAs were from [107]; see Datafile S4.